The use of 201T1 has been proposed for the differential diagnosis of lymphomas and non-neoplastic brain masses in AIDS patients. The aim of this study was to assess the diagnostic accuracy of three different semi-quantitative methods for the analysis of 201T1 SPET brain images in individuals with AIDS and brain lesions. Thirty-seven AIDS patients with contrast-enhancing brain lesions underwent 201T1 SPET. Three different lesion-to-background uptake indices were calculated: (1) small lesion/large background (SL/LB; i.e. counts in a 3 x 3 pixel ROI in the lesion/counts in a 7 x 7 pixel ROI in the contralateral healthy hemisphere); (2) small lesion/multiple small background (SL/MSL; i.e. counts in a 3 x 3 pixel ROI in the lesion/average counts of ten 3 x 3 ROIs over the highest background pixel values); (3) large lesion/large background (LL/LB; i.e. counts in an elliptic ROI in the lesion/counts in a contralateral mirrored ROI). Data analysis included a ROC curve analysis to identify the cut-off value corresponding to the highest accuracy value, and an analysis of the predictive value to classify the patients in three categories (high, intermediate and low risk of lymphoma). The greatest accuracy (71%) was achieved with the LL/LB method of analysis. Using this method, 62% of patients could be classified as either having lymphoma or not, whereas 38% could not be classified. LL/LB values > or = 2.9 are suggestive of the presence of lymphomas, whereas values < or = 2 are highly predictive of the presence of a lesion other than lymphoma. However, LL/LB values between 2 and 3 are not diagnostic and adjunctive tests should be carried out. In conclusion, 201T1 SPET was an adequate diagnostic tool in approximately 70% of the cases in this study.